On Fri, 30 Jan 2026 23:36:31 +0000, Seymour J Metz wrote:
>Or accept a trailing colon and allow continuation, in honor of PL/I syntax.
> ...
The design is misguided. It should be:
First: Completely resolve continuations.
Then: Perform remaining lexical analysis.
JCL has a similar problem in that symbols must not be
continued, making generating JCL by a program needlessly
complex.
I once tried to edge-test JCL parsing by using a large
number of symbols all of which evaluated to the null
string. I gave up before encountering an (undocumented)
limit on the number of continuation lines.
Completely performing concatenation first might require
prohibitively large working storage. This could be
avoided by making continuation resolution a coroutine
of remaining lexical analysis.
>________________________________________
>From: Richard Zierdt
>Sent: Friday, January 30, 2026 3:35 PM
> ...
>Long_Field_Name_Defined_On_Next_Line + <-- column 72
> DC CL8'APPLES'
>returns
>ASMA142E Operation code not complete on first record
>
>Long_Field_Name_Defined_On_Next_Line DC + <-- column 72
> CL8'APPLES'
>works.
--
gil
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
